Competencies and Education

Some state actors violate rule-of-law principles because they do not understand them or lack the skills or resources to adhere to them. In Liberia, for example, where the legal system incorporates elements of US common law, inadequately trained judges have often issued apparently arbitrary rulings because they do not understand the distinction between precedents based on US common law and those based on US statutes that do not apply in Liberia (Lubkemann, Isser, and Banks III Reference Lubkemann, Isser, Banks and Isser2011, 205). This is a problem of competencies, which UN missions attempt to solve through training, technical support, and other forms of education, broadly defined.

Education is one of the UN’s most widely used mechanisms for socializing state actors to align their behavior with rule-of-law (and other) principles (Howard Reference Howard2019). In the Central African Republic (CAR), for example, BINUCA helped the Ministry of Justice revise the Criminal and Criminal Procedure Codes in accordance with international law, then hosted seminars to improve understanding of the Codes among judges, lawyers, law school professors, court clerks, police officers, and gendarmes. The mission also sought to educate both citizens and state actors on their new legal rights and responsibilities by distributing some 3,000 hard copies of the Codes nationwide (UN Criminal Law and Judicial Advisory Service 2013).

Incentives and Inducement

Other state actors have the skills and information necessary to adhere to rule-of-law principles, but violate them because they benefit from doing so. In the Democratic Republic of the Congo (DRC), for example, police officers derive much of their income from institutionalized illicit transfers, including extortion of civilians. This is a problem of incentives, which UN missions attempt to solve through inducement. Like other international organizations, UN missions “have the advantage of resources and leverage over weak or developing states they seek to convert to their normative convictions” (Finnemore and Sikkink Reference Finnemore and Sikkink1998, 900). They use these resources and leverage as mechanisms of inducement.

In particular, UN missions are conduits for funding and other forms of material support from the international community, which are often tied to the implementation of particular reforms. In Côte d’Ivoire, for example, UNOCI raised millions of dollars to rehabilitate the courts and prisons while also creating an office within the Ministry of Justice to vet and monitor judges (UN Department of Peacekeeping Operations 2010, 10). In Côte d’Ivoire and elsewhere, state actors who benefit from the increased capacity that UN missions provide may be willing to undertake other, potentially more controversial reforms in order to secure additional material support. Like education, inducement can also enable state officials to perform their duties more effectively (and potentially more consistently with rule-of-law principles)—for example by paying the salaries of civil servants or by providing police officers with vehicles and computers.

Beliefs and Persuasion

Other state actors violate rule-of-law principles because they do not believe those principles are acceptable, appropriate, or legitimate. In Mali, for example, justice sector personnel often refuse to investigate abuses by state security forces because they believe due process protections should not extend to suspected enemies of the state (Human Rights Watch 2018, 359–62). This is a problem of beliefs, which UN missions attempt to solve through persuasion (Howard Reference Howard2019, 42). International organizations “are rarely able to coerce agreement to a norm—they must persuade” (Finnemore and Sikkink Reference Finnemore and Sikkink1998, 900). In South Sudan, for example, UNMISS established liaison officers in prisons and mobile courts in order to “change the mindset of judges, prosecutors, and prison officials regarding prolonged and arbitrary detention” (UN Department of Peacekeeping Operations 2014, 6, my emphasis). These initiatives are designed to “take what is seen as natural or appropriate and convert it into something perceived as wrong” (Finnemore and Sikkink Reference Finnemore and Sikkink1998, 900).

Accountability and Oversight

Finally, some state actors violate rule-of-law principles because they do not expect to be caught or punished for doing so. In Côte d’Ivoire, for example, a breakdown in civilian control over the police, army, and gendarmerie fostered impunity as “renegade forces” created “pockets of disobedience and insubordination” (Outtara Reference Outtara, Bryden, N’Diaye and Olonisakin2008, 82, 89). This is a problem of accountability, which UN missions attempt to solve through monitoring, surveillance, and other forms of oversight.

Oversight constitutes a mechanism of “accountability politics” whereby “increased attention, followed by changes in discursive positions” make state actors more “vulnerable” to reform (Keck and Sikkink Reference Keck and Sikkink1998, 26). When combined with arm-twisting or naming and shaming, oversight can also serve as a mechanism of coercion that does not rise to the level of physical force (Howard Reference Howard2019, 141). In Namibia, for example, UNTAG police officers trained their local counterparts in “policing by consent” standards, then monitored compliance with those standards by conducting joint foot and vehicle patrols, codeploying in police stations, and reporting acts of misconduct up the chain of command (Howard Reference Howard2008, 75–6).

Personnel: Capacity and Composition

UN missions vary in their capacity to pursue rule-of-law reform. While capacity has many dimensions, most scholars proxy for it using personnel numbers (Hultman, Kathman, and Shannon Reference Hultman, Kathman and Shannon2013; Reference Hultman, Kathman and Shannon2014; Kathman et al. Reference Kathman, Beardsley, Gizelis, Olsson, Bove, Ruggeri and Zwetsloot2017). Personnel numbers matter for rule-of-law promotion for at least two reasons. First, efforts to reform the four links in the justice chain are labor intensive, especially when undertaken simultaneously. The larger the mission, the more human resources it can devote to these tasks. Second, reform is more likely to succeed when state actors believe the international community is committed to the process. Personnel numbers are arguably the “strongest observable signal of the UN’s resolve” (Ruggeri, Gizelis, and Dorussen Reference Ruggeri, Gizelis and Dorussen2012, 388).

But not all UN personnel contribute equally to the UN’s rule-of-law agenda. Most quantitative peacekeeping studies focus exclusively on uniformed personnel (Hultman, Kathman, and Shannon Reference Hultman, Kathman and Shannon2013; Reference Hultman, Kathman and Shannon2014; Ruggeri, Gizelis, and Dorussen Reference Ruggeri, Gizelis and Dorussen2012). Uniformed personnel can have important indirect effects on the rule of law by maintaining stability, neutralizing spoilers, and creating the conditions under which more direct attempts at reform can occur. But rule-of-law principles are unlikely to become entrenched through the presence of uniformed personnel alone. Arguably more important are the UN’s civilian personnel, who are almost solely responsible for court, prison, and legal reform, and who share responsibility for police reform with UN police officers.

UN civilian and police contingents perform “critical functions” that cannot be easily outsourced to their military counterparts, including “developing national rule of law plans and strategies, coordinating national and international stakeholders, advising on justice matters, and providing technical expertise” (Sannerholm et al. Reference Sannerholm, Möller, Simion and Hallonsten2012, 8). While UN troops and military observers may be more effective at separating belligerents, UN civilian and police personnel are likely to be more effective at educating, inducing, persuading, and overseeing host state police forces, courts, prisons, and legislatures. As the UN itself has acknowledged, these initiatives require significant “civilian capacity” (HIPPO 2015, 41).

Rule-of-law reform is also more likely to be effective when UN missions assign civilian and police personnel specifically to rule-of-law-related tasks. Rule-of-law reform demands “dedicated capacities” (UN Secretary-General 2006, 1–2, 8) and “specialized expertise” (HIPPO 2015, 41). Assigning personnel specifically to rule-of-law-related tasks is especially important in missions with complex, multidimensional mandates, where potentially competing goals can undermine the mission’s ability to pursue any of them effectively (Blair, di Salvatore, and Smidt Reference Blair, di Salvatore and Smidt2020). By assigning personnel specifically to rule-of-law-related tasks, missions help ensure that rule-of-law promotion remains a priority even as mandated tasks proliferate. My first two hypotheses follow from these observations:

Activities: Mandate Implementation and Engagement with Host States

Even UN missions with similar mandates may implement those mandates in different ways (Kathman et al. Reference Kathman, Beardsley, Gizelis, Olsson, Bove, Ruggeri and Zwetsloot2017), and even missions with similar capacities and compositions may vary in the activities they pursue in the field (Smidt Reference Smidt2016). UN missions exercise discretion over these tactical and operational decisions, allowing them to execute their mandates in unintended and unanticipated ways (Barnett and Finnemore Reference Barnett and Finnemore1999, 699). As Blair, di Salvatore, and Smidt (Reference Blair, di Salvatore and Smidt2020) show, the extent to which a given UN mission even attempts to fulfill the tasks assigned to it in its mandate can fluctuate dramatically from one year to the next. These fluctuations constitute an important but understudied source of peacekeeping success or failure (Lipson Reference Lipson2010). I argue that establishing the rule of law requires reforming the four links in the justice chain. Whether peacekeepers succeed in instituting these reforms depends on whether they actually attempt to do so.Footnote ³

Even peacekeeping operations that pursue the same combination of police, court, prison, and legal reform may do so in different ways. I argue that UN missions are most successful when they pursue rule-of-law reform through a combination of education, inducement, persuasion, and oversight. Importantly, all of these mechanisms involve engaging host states in the reform process. But as Dorussen and Gizelis (Reference Dorussen and Gizelis2013, 6) note, the nature of peacekeepers’ “specific involvement” with host states “varies widely from technical assistance and monitoring to direct line authority or even direct implementation”—that is, performing mandated tasks without host state involvement. In CAR, for example, MINUSCA was authorized to use “Urgent Temporary Measures” to maintain law and order in areas where the police and courts were absent or dysfunctional. Other missions have similarly bypassed host states in the process of reform.

But this approach is risky, for at least five reasons. First, UN missions that bypass host states may weaken the capacity of the very same institutions they are ostensibly mandated to strengthen (Lake Reference Lake, Hoddie and Hartzell2010). Second, bypassing host states may provoke backlash from state actors, who may “contest peacekeeping actions that replace rather than improve their regulatory capacity” (Dorussen and Gizelis Reference Dorussen and Gizelis2013, 6). Third and related, bypassing may create incentives for state actors to erect “parallel structures of government” that later pose “serious obstacles” to the rule of law (Beauvais Reference Beauvais2001, 1117).

Fourth, UN missions that bypass host states may forgo opportunities to influence the beliefs and behaviors of government officials and the structure of government institutions. This may prolong the reform process, thus allowing state dysfunction and predation to persist. Fifth, bypassing host states may increase the risk that reforms prove inappropriate or inadequate for the contexts in which they are implemented. For all of these reasons, the UN itself has advocated for “high-level political engagement in support of national [reform] efforts” (HIPPO 2015, x) and has urged peacekeepers to “identify, support, and empower domestic reform constituencies” rather than circumvent them (UN Secretary-General 2004, 8). My third and fourth hypotheses therefore predict the following:

Timing

Finally, UN missions vary in the timing of their deployments and the activities they pursue. In particular, peacekeepers are increasingly expected to advance the UN’s rule-of-law agenda even in settings where civil war is ongoing. Indeed, as we will see, UN missions are actually more likely to pursue rule-of-law reform during periods of conflict than periods of peace. But reform is also less likely to succeed amidst ongoing civil war. During periods of conflict, UN missions risk being perceived not as impartial forces but as warring parties, thus exposing themselves to threats of violence (Karlsrud Reference Karlsrud2015). Rule-of-law reform is less likely to be effective when UN personnel are preoccupied with protecting themselves, and civilians, from imminent physical harm.

Moreover, as critics both inside and outside the UN system have warned, UN missions are often “poorly suited to these operating environments,” making their mandates “less achievable” (HIPPO 2015, x, 47). Not coincidentally, previous studies have found that peacekeeping operations are less successful during periods of conflict than periods of peace (Doyle and Sambanis Reference Doyle and Sambanis2006; Gilligan and Sergenti Reference Gilligan and Sergenti2008). Challenges that UN missions struggle to overcome in peacetime become even more intractable in the “fog of war,” including culture clashes, turf battles, and coordination problems both between and within civilian and uniformed contingents (Blair, di Salvatore, and Smidt Reference Blair, di Salvatore and Smidt2020), which comprise personnel from diverse settings, with disparate levels of training and expertise. This is especially problematic for the UN’s rule-of-law agenda, which requires carefully coordinating efforts to reform the four links in the justice chain simultaneously.

Perhaps most important, state actors are unlikely to have the will or capacity to participate in the reform process if they are busy fighting rebel groups. Indeed, state actors may reject some reforms out of hand if they are perceived as impediments to counterinsurgency (Blair, di Salvatore, and Smidt Reference Blair, di Salvatore and Smidt2020)—restrictions on pretrial detention, for example, or on the use of force by police officers. Peace processes create “windows of opportunity” for reform (Call Reference Call2007, 11), and the UN itself has identified the first few years after civil war termination as the most “opportune moment to work for change” (Carlson Reference Carlson2006, 2). This window of opportunity is likely to be narrower when there is no peace to keep, and even reforms that begin during conflict are only likely to become institutionalized after the fighting stops. My fifth hypothesis follows from these propositions:

Operationalizing UN Personnel and Activities

My theory predicts that the efficacy of UN rule-of-law promotion should vary as a function of UN personnel, activities, and timing. I operationalize UN personnel and activities using four sources of data. First, I use Providing for Peacekeeping (P4P) data on the number of uniformed personnel deployed to each UN mission in Africa since 1990.Footnote ⁴ While P4P distinguishes between UN troops, police officers, and military observers, different categories of uniformed personnel tend to be highly correlated with one another, and I combine them into a single count. I discuss the issue of multicollinearity in further detail in Appendix G, and in Appendix J I disaggregate military and police personnel, as the latter tend to be more intimately involved in rule-of-law reform than the former. Figure A.1 plots the distribution of uniformed personnel across Africa.

Often it is the UN’s civilian components that are most directly responsible for advancing the UN’s rule-of-law agenda. I use annual budget performance reports prepared by the UN’s Advisory Committee on Administrative and Budgetary Questions (ACABQ) to build what is, to my knowledge, the first dataset of civilian contributions to UN peacekeeping operations. ACABQ reports distinguish four categories of civilian personnel: international staff, national staff, government-provided personnel (GPPs), and UN Volunteers (UNVs). To avoid multicollinearity, I again collapse these categories into a single count. In Appendix K, I disaggregate international and national civilian staff, as the former tend to be better paid and enjoy higher status within UN missions than the latter. Figure A.2 plots the distribution of civilian personnel across the nine African countries for which budget performance reports are available.Footnote ⁵

Even within the civilian components of UN missions, different personnel play different roles. I use the UN Secretary-General’s annual budget requests to code the number of UN personnel assigned specifically to rule-of-law-related tasks. Budget requests are unique in specifying the number of personnel assigned to each dimension of a mission’s mandate, as the example in Figure A.3 illustrates. But budget requests are also limited in two ways. First, in most cases budget requests only begin drawing distinctions between mandate components in 2003. Second, budget requests only capture the number of personnel approved by the UN General Assembly. As I discuss in Appendix C, however, the correlation between approved and actual personnel numbers is generally very close to 1.00, suggesting that the former is a useful proxy for the latter. I discuss my coding rules and missing data issues in detail in Appendix C. To my knowledge, this is the first dataset to disaggregate UN personnel by mandate component. Figure A.4 plots the distribution of UN personnel assigned specifically to rule-of-law-related tasks across the nine countries for which budget requests are available.

Finally, in collaboration with Hannah Smidt, I use UN Secretary-General progress reports to code the extent to which UN missions actually pursue rule-of-law-related activities on the ground. Progress reports are typically published four to seven times per year. We use them to code 37 different activities from 1989 to 2016, ranging from demining to voter education to humanitarian relief, and encompassing virtually the entire universe of activities that UN missions implemented during these years. Following my theoretical framework, I focus on activities targeting the four links in the justice chain—police forces, courts, prisons, and laws—through four mechanisms of engagement with host states: education, inducement, persuasion, and oversight. I also code implementation of rule-of-law reform without host state involvement. In Appendix D, I discuss coding rules, intercoder reliability, and limitations—in particular, the possibility that progress reports may be selective and may emphasize activities that the UN believes best reflect its achievements. Our Peacekeeping Activities Dataset (PACT) extends earlier efforts by Ruggeri, Gizelis, and Dorussen (Reference Ruggeri, Gizelis and Dorussen2011) and Smidt (Reference Smidt2016). PACT is more comprehensive than its predecessors, substantively, geographically, and temporally. I compare PACT to these other datasets in Appendix D.

Table A.1 summarizes the prevalence of police, court, prison, and legal reform in the PACT dataset. Table A.1 also summarizes the prevalence of rule-of-law-related education, inducement, persuasion, and oversight. The UN typically pursues all four categories of reform simultaneously, using a combination of all four mechanisms of engagement, and the pairwise correlations between rule-of-law-related activities are very high—always between 0.68 and 0.91. These correlations are important in themselves, as they suggest the UN has succeeded in pursuing a relatively “balanced and holistic” approach to rule-of-law reform (Carlson Reference Carlson2006, 2). Simultaneity is, however, disadvantageous for my analysis, as multicollinearity prevents me from estimating the effects of each activity conditional on the others.

Operationalizing the Rule of Law

To operationalize the rule of law, I rely primarily on two widely used indices from the World Bank and Freedom House. Freedom House’s rule-of-law index is based on reports from country and subject matter experts. The World Bank’s index aggregates data from 32 public, private, and nonprofit sources, including Freedom House. The World Bank’s index is available for most countries in the world beginning in 1996. The Freedom House index is available for most countries beginning in 2003. The World Bank’s index is standardized to have 0 mean and unit standard deviation globally. To facilitate comparison to other regions, I do not restandardize to the countries in my sample. Freedom House’s index is scaled from 0 to 16. Table A.2 reports descriptive statistics, and Figures A.5 and A.6 plot trends in the two indices over time.

The World Bank defines the rule of law as “the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence.”Footnote ⁶ This definition is both broader and narrower than the UN’s, encompassing some secondary dimensions of the rule of law (the incidence of crime, for example) while ignoring some primary ones (for example, the extent to which laws are publicly promulgated). Nonetheless, the World Bank’s definition echoes the UN’s and captures several key components of the rule of law, including the quality of state security and justice provision and the extent to which government officials adhere to legal and constitutional constraints. Among the variables in the World Bank’s index are 16 proxies for the fairness, efficacy, and reputation of the courts; five indicators for the reliability of the police; and four measures of the independence of the judiciary. A list of sources included in the index is in Appendix A.

Freedom House’s definition of the rule of law comprises four parts: independence of the judiciary, respect for due process, protection from the “illegitimate use of physical force and freedom from war and insurgencies,” and equality of access to justice.Footnote ⁷ While also somewhat broader than the UN’s, Freedom House’s definition incorporates country experts’ responses to a battery of questions that correspond closely to the UN’s understanding of the rule of law. A list of these questions is in Appendix B. Importantly, while Freedom House includes “freedom from war and insurgencies” as one component of the rule of law, scores on the index vary widely even during periods of conflict.Footnote ⁸

The rule of law is a multidimensional concept, and the World Bank and Freedom House indices are broad enough to capture multiple dimensions simultaneously. As a robustness check, Appendices L through N replicate my analysis using four narrower indices designed to capture more specific rule-of-law principles. One key principle is the extent to which laws constrain the powers of the executive. Appendix L operationalizes this principle using the Polity IV “executive constraints” index, which captures “the extent of institutionalized constraints on the decision-making powers of chief executives.”Footnote ⁹ Appendix M operationalizes this same principle using the Varieties of Democracy (V-Dem) “judicial constraints on the executive” index, which captures the extent to which the executive respects the constitution and complies with the courts.

A second crucial principle is the extent to which laws are enforced equally and transparently. Appendix M uses V-Dem’s “equality before the law and individual liberty” index to operationalize this principle. A third important principle is the extent to which the judiciary is independent from the other branches of government, especially the executive. Appendix N operationalizes this principle using the Linzer and Staton (Reference Linzer and Staton2015) index of judicial independence. (The V-Dem judicial constraints index partially captures this principle as well.) Results using these four indices are similar to those reported below. Appendix O provides a correlation matrix for the six indices. Not surprisingly, they are all positively correlated, but the correlations are not so high as to suggest they are merely substitutes for one another.

ESTIMATION AND IDENTIFICATION

Changes in UN personnel and activities may be correlated with changing conditions on the ground, which, in turn, may be correlated with the rule of law. It is not clear whether we should expect selection effects of this sort to bias my estimates towards or away from the null. While the UN has failed to intervene in some of the world’s most intractable civil wars—Syria, for example—there is ample quantitative evidence that, in general, peacekeepers are more likely to deploy to the “hardest” cases, attenuating the correlation between UN presence and peace (Doyle and Sambanis Reference Doyle and Sambanis2006; Fortna Reference Fortna2008a; Gilligan and Stedman Reference Gilligan and Stedman2003). By the same logic, if the number of personnel deployed to a UN mission increases during the hardest moments for rule-of-law reform, then we should expect the relationship between UN personnel and the rule of law to be biased towards the null. Similarly, if UN missions are more likely to pursue rule-of-law-related activities in the hardest cases, then we should expect the relationship between UN activities and the rule of law to be biased towards the null as well.

I explore the extent to which UN missions favor the hard cases for rule-of-law reform in two ways. First, following my theoretical framework, rule-of-law reform should be less likely to succeed during periods of conflict than periods of peace. Yet in my sample, I find that UN missions are more rather than less likely to pursue rule-of-law-related activities during periods of conflict. The UN missions in my sample pursue rule-of-law-related activities in 15% of all country-years of conflict, compared with just 9% of all country-years of peace. This discrepancy holds across three of the four categories of rule-of-law reform (police, court, and prison reform, but not legal reform) and across all four mechanisms of engagement with host states (education, inducement, persuasion, and oversight). In other words, peacekeepers appear to be more likely to pursue rule-of-law-related activities at moments when those activities are less likely to succeed. This should attenuate the relationship between rule-of-law-related activities and the rule of law.

UN missions tend to expand in size during periods of conflict as well. The number of uniformed personnel in my sample is 28% larger during country-months of conflict (2,806 on average) than country-months of peace (2,197). Nor is this discrepancy specific to uniformed personnel: the number of civilian personnel is more than twice as large during country-months of conflict (799 on average) than country-months of peace (307). This should attenuate the relationship between UN personnel and the rule of law. While the number of personnel assigned specifically to rule-of-law-related tasks tends to be larger during periods of peace than periods of conflict, the discrepancy is small, with an average of 639 personnel assigned specifically to rule-of-law-related tasks during country-months of conflict, compared with 732 during country-months of peace.

Second, and again following my theoretical framework, rule-of-law reform should be less likely to succeed when threats to the rule of law are more severe. Yet as I show in Appendix I, the UN missions in my sample are equally or more likely to expand in size when threats to the rule of law are more severe and equally or more likely to pursue rule-of-law-related activities.Footnote ¹⁰ Again, if anything, this should bias my results towards the null.Footnote ¹¹ This should also help mitigate concerns about trending—that is, the possibility that any positive correlation I observe between UN presence and the rule of law is an artifact of improvements in the rule of law that occur endogenously as peace consolidates and UN missions expand or deterioration that occurs endogenously as missions contract. Trending of this sort does not appear to be a cause for concern, at least in these countries and years.

I address three additional threats to causal inference in the appendix. First, it is possible that the effects of UN personnel and activities may spill over from one country to another, biasing my results. My use of country fixed effects should mitigate this risk, as I discuss in Appendix H. Second, even if UN missions are more likely to pursue rule-of-law-related activities when threats to the rule of law are more severe, those activities may require bypassing rather than engaging the host state—for example, if the local police force is implicated in ongoing human rights violations. In this case, the correlation between bypassing and the rule of law would be biased towards the null, while the correlation between engagement and the rule of law would be biased away from it. In Appendix I, I show that this is unlikely, as UN missions in my sample are actually more likely to engage host states when threats to the rule of law are more severe.

Third and somewhat less plausibly, it is possible that any positive association I observe between UN presence and the rule of law is not a result of police, court, prison, or legal reform but is rather a byproduct of other activities that UN missions pursue at the same time—for example, promotion of public health or economic growth. Empirically assessing this possibility is a challenge, as UN missions typically pursue multiple political, social, and economic reforms simultaneously. Nonetheless, as a placebo test, in Appendix P I use PACT to identify a set of activities that are not likely to affect the rule of law and only weakly correlated with activities that are. I focus in particular on the imposition and enforcement of UN arms embargoes. I find that the correlation between arms embargo-related activities and the rule of law is generally null or negative. This is consistent with the (rather intuitive) proposition that rule-of-law-related activities are more likely than other types of activities to promote the rule of law.

I attempt to minimize any remaining bias by using country fixed effects to eliminate time-invariant confounders and by controlling for some of the most problematic time-varying confounders: population, GDP per capita, foreign aid, infant mortality, quality of democracy, and an indicator for any violence against the regime.Footnote ¹² Importantly, my use of country fixed effects implies that I am estimating off of variation that arises within rather than between countries. This is advantageous for causal inference but potentially disadvantageous for the scope of my analysis, as I cannot compare countries that hosted peacekeeping operations with those that did not. (There is no variation in any of my independent variables in countries that did not host at least one peacekeeping operation.) As a robustness check, in Appendix S I report results from a random effects estimator that exploits variation both within and across countries, and which therefore incorporates countries that did not host peacekeeping operations. My results are substantively similar to those reported here.

Ideally I could test the relationship between UN presence and the rule of law both while UN missions are on the ground and after they withdraw. Unfortunately, the length of most UN mandates precludes this option. Instead I run my analyses both during civil war and one, two, and three years after civil war termination, using the Uppsala Conflict Data Program’s (UCDP) armed conflict dataset to code civil war onset and termination. In some countries, these specifications do in fact encompass the period after UN withdrawal. For example, the Burundian civil war ended in 2005, one year after the UN deployed and one year before it withdrew. Similarly, the most recent Chadian civil war ended in 2010, the same year the UN drew down. In these cases my analysis captures the entire trajectory from deployment to implementation to withdrawal. For an approach similar to mine, see Fortna (Reference Fortna, Jarstad and Sisk2008b).

Finally, ideally I could compare UN rule-of-law-related activities with similar activities undertaken by other domestic and international organizations—the World Bank and the Carter Center, for example. While PACT contains data on these activities, UN Secretary-General progress reports usually focus on activities implemented in collaboration with UN missions. As a result, the correlation between UN and non-UN rule-of-law promotion in PACT is close to 1.00. This is not as limiting as it may seem, as many international organizations coordinate with the UN when designing and implementing rule-of-law programs. Many of these programs appear in the UN’s progress reports, and those that do not likely fall under the purview of relatively small domestic NGOs. Still, some ambiguity remains, and my analysis is best interpreted as capturing UN rule-of-law reform with assistance from other international organizations.